Pub Date : 2022-02-12DOI: 10.1109/AISP53593.2022.9760596
K. Kumar, J. Harikiran, B. S. Chandana
Human activity recognition is an extensively researched topic in the field of computer vision. Recognizing human activities without revealing a person’s identity is one such use case. To solve this, we propose a practical method for human activity recognition (HAR) while maintaining anonymity. It captures and distributes data from a variety of sources while respecting the privacy of the individuals concerned. At the core of our approach is (DBN-RGMAA) based on deep neural networks, which are not only more accurate but can also be deployed in real-time video surveillance systems. Hence, this work presents a deep learning-based scheme for privacy-preserving human activities. Initially, for extracting the features from raw video data, a Deep Belief Network (DBN) is used. To increase the HAR identification rate, Hybrid Deep Fuzzy Hashing Algorithm (HDFHA) is employed to capture dependencies between two actions. Finally, the privacy model enhances the privacy of humans while permitting a highly accurate approach towards action recognition by the Recursive Genetic Micro-Aggregation Approach (RGMAA). The implementation is executed and the performances are evaluated by Accuracy, Precision, Recall, and F1 Score. A dataset named HMDB51 is used for empirical study. Our experiments using the Python data science platform reveal that the OPA-PPAR outperforms existing methods.
{"title":"Human Activity Recognition with Privacy Preserving using Deep Learning Algorithms","authors":"K. Kumar, J. Harikiran, B. S. Chandana","doi":"10.1109/AISP53593.2022.9760596","DOIUrl":"https://doi.org/10.1109/AISP53593.2022.9760596","url":null,"abstract":"Human activity recognition is an extensively researched topic in the field of computer vision. Recognizing human activities without revealing a person’s identity is one such use case. To solve this, we propose a practical method for human activity recognition (HAR) while maintaining anonymity. It captures and distributes data from a variety of sources while respecting the privacy of the individuals concerned. At the core of our approach is (DBN-RGMAA) based on deep neural networks, which are not only more accurate but can also be deployed in real-time video surveillance systems. Hence, this work presents a deep learning-based scheme for privacy-preserving human activities. Initially, for extracting the features from raw video data, a Deep Belief Network (DBN) is used. To increase the HAR identification rate, Hybrid Deep Fuzzy Hashing Algorithm (HDFHA) is employed to capture dependencies between two actions. Finally, the privacy model enhances the privacy of humans while permitting a highly accurate approach towards action recognition by the Recursive Genetic Micro-Aggregation Approach (RGMAA). The implementation is executed and the performances are evaluated by Accuracy, Precision, Recall, and F1 Score. A dataset named HMDB51 is used for empirical study. Our experiments using the Python data science platform reveal that the OPA-PPAR outperforms existing methods.","PeriodicalId":6793,"journal":{"name":"2022 2nd International Conference on Artificial Intelligence and Signal Processing (AISP)","volume":"64 3 1","pages":"1-8"},"PeriodicalIF":0.0,"publicationDate":"2022-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78042055","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-02-12DOI: 10.1109/AISP53593.2022.9760670
Amena Qadri Syed, J. Jothi, K. Anusree
Early crack identification of civil structures is an essential task to prolong the life of the structures and to promise public safety. This research aims to develop an automated crack identification system using deep learning models and the SDNET2018 dataset. Image augmentation is applied to overcome the effect of unbalanced data. Deep pre-trained models like VGG16, InceptionV3, ResNet-50, ResNet-101 and ResNet-152 are trained and tested using the cracked and uncracked images of decks and pavements from the dataset. The experimental results show that the classification models obtained using transfer learning on the cracked and non-cracked pavement and deck image dataset have accuracy values of 70.59%, 60.31%71.93%, 75.40%, and 74.77% for VGG-16, Inception V3, ResNet-50, ResNet-101, and Resnet-152 pretrained models respectively.
{"title":"Crack identification from concrete structure images using deep transfer learning","authors":"Amena Qadri Syed, J. Jothi, K. Anusree","doi":"10.1109/AISP53593.2022.9760670","DOIUrl":"https://doi.org/10.1109/AISP53593.2022.9760670","url":null,"abstract":"Early crack identification of civil structures is an essential task to prolong the life of the structures and to promise public safety. This research aims to develop an automated crack identification system using deep learning models and the SDNET2018 dataset. Image augmentation is applied to overcome the effect of unbalanced data. Deep pre-trained models like VGG16, InceptionV3, ResNet-50, ResNet-101 and ResNet-152 are trained and tested using the cracked and uncracked images of decks and pavements from the dataset. The experimental results show that the classification models obtained using transfer learning on the cracked and non-cracked pavement and deck image dataset have accuracy values of 70.59%, 60.31%71.93%, 75.40%, and 74.77% for VGG-16, Inception V3, ResNet-50, ResNet-101, and Resnet-152 pretrained models respectively.","PeriodicalId":6793,"journal":{"name":"2022 2nd International Conference on Artificial Intelligence and Signal Processing (AISP)","volume":"7 1","pages":"1-7"},"PeriodicalIF":0.0,"publicationDate":"2022-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"72972780","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-02-12DOI: 10.1109/AISP53593.2022.9760543
M. Karuna, S. R. Guntur
The recognition of hand movements using surface electromyography (sEMG) and a machine learning technique is becoming increasingly significant to control a prosthetic hand in a rehabilitation facility for people who have had their hands amputated in order to regain lost capability. However, in real life, controlling a prosthetic hand utilizing non-invasive methods is still a challenge. Existing research results are limited and not meeting the needs of amputee. The objective of this work is to fulfill the gap by proposing empirical mode decomposition (EMD) based machine learning (ML)classifier to recognize hand movements of the Ninapro dataset, this benchmark standard is used to evaluate four classifiers by comparing the performance accuracy results. The outcome of this work is better movement recognition achieved using one of the four distinct classifiers.
{"title":"Classification of Hand Movements via EMG using Machine Learning Methods for Prosthesis","authors":"M. Karuna, S. R. Guntur","doi":"10.1109/AISP53593.2022.9760543","DOIUrl":"https://doi.org/10.1109/AISP53593.2022.9760543","url":null,"abstract":"The recognition of hand movements using surface electromyography (sEMG) and a machine learning technique is becoming increasingly significant to control a prosthetic hand in a rehabilitation facility for people who have had their hands amputated in order to regain lost capability. However, in real life, controlling a prosthetic hand utilizing non-invasive methods is still a challenge. Existing research results are limited and not meeting the needs of amputee. The objective of this work is to fulfill the gap by proposing empirical mode decomposition (EMD) based machine learning (ML)classifier to recognize hand movements of the Ninapro dataset, this benchmark standard is used to evaluate four classifiers by comparing the performance accuracy results. The outcome of this work is better movement recognition achieved using one of the four distinct classifiers.","PeriodicalId":6793,"journal":{"name":"2022 2nd International Conference on Artificial Intelligence and Signal Processing (AISP)","volume":"1 1","pages":"1-4"},"PeriodicalIF":0.0,"publicationDate":"2022-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"80070146","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-02-12DOI: 10.1109/AISP53593.2022.9760564
Rashi Kumar, V. Sahula
Sanskrit is a low resource language for which large parallel data sets are not available. Large parallel data sets are required for Machine Translation. Cross-Lingual word embedding helps to learn the meaning of words across languages in a shared vector space. In the present work, we propose a translation technique between Sanskrit and Hindi words without a parallel corpus-base. Here, fastText pre-trained word embedding for Sanskrit and Hindi are used and are aligned in the same vector space using Singular Value Decomposition and a Quasi bilingual dictionary. A Quasi bilingual dictionary is generated from similar character string words in the monolingual word embeddings of both languages. Translations for the test dictionary are evaluated on the various retrieval methods e.g. Nearest neighbor, Inverted Sofmax approach, and Cross-domain Similarity Local Scaling, in order to address the issue of hubness that arises due to the high dimensional space of the vector embeddings. The results are compared with the other Unsupervised approaches at 1, 10, and 20 neighbors. While computing the Cosine similarity, we observed that the similarity between the expected and the translated target words is either close to unity or equal to unity for the cases that were even not included in the Quasi bilingual dictionary that was used to generate the orthogonal mapping. A test dictionary was developed from the Wikipedia Sanskrit-Hindi Shabdkosh to test the translation accuracy of the system. The proposed method is being extended for sentence translation.
{"title":"Word Translation using Cross-Lingual Word Embedding: Case of Sanskrit to Hindi Translation","authors":"Rashi Kumar, V. Sahula","doi":"10.1109/AISP53593.2022.9760564","DOIUrl":"https://doi.org/10.1109/AISP53593.2022.9760564","url":null,"abstract":"Sanskrit is a low resource language for which large parallel data sets are not available. Large parallel data sets are required for Machine Translation. Cross-Lingual word embedding helps to learn the meaning of words across languages in a shared vector space. In the present work, we propose a translation technique between Sanskrit and Hindi words without a parallel corpus-base. Here, fastText pre-trained word embedding for Sanskrit and Hindi are used and are aligned in the same vector space using Singular Value Decomposition and a Quasi bilingual dictionary. A Quasi bilingual dictionary is generated from similar character string words in the monolingual word embeddings of both languages. Translations for the test dictionary are evaluated on the various retrieval methods e.g. Nearest neighbor, Inverted Sofmax approach, and Cross-domain Similarity Local Scaling, in order to address the issue of hubness that arises due to the high dimensional space of the vector embeddings. The results are compared with the other Unsupervised approaches at 1, 10, and 20 neighbors. While computing the Cosine similarity, we observed that the similarity between the expected and the translated target words is either close to unity or equal to unity for the cases that were even not included in the Quasi bilingual dictionary that was used to generate the orthogonal mapping. A test dictionary was developed from the Wikipedia Sanskrit-Hindi Shabdkosh to test the translation accuracy of the system. The proposed method is being extended for sentence translation.","PeriodicalId":6793,"journal":{"name":"2022 2nd International Conference on Artificial Intelligence and Signal Processing (AISP)","volume":"10 1","pages":"1-7"},"PeriodicalIF":0.0,"publicationDate":"2022-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"75491016","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-02-12DOI: 10.1109/AISP53593.2022.9760625
Biji Rose, B. Arunadevi
In a wireless radio environment of cognitive radio, spotting of vacant spectrum of Primary user demands more efficient technique. The edge detection of sub-bands of the received signal spectrum is one such efficient technique of spectrum sensing achieved by Discrete Wavelet Transform (DWT). In low noise variance, the DWT based technique has a better detection performance, but as noise variance increases, the performance degrades. In this paper, blind energy detection spectrum-sensing approach is proposed with Multiscale DWT. Here depending on the noise variance two modified forms of DWT are proposed. When noise variance is less DWT Modulus Maxima (DWTMM) and for high noise variance DWT Moving window ESPIT Method (DWTMEM). The simulation of the proposed algorithm, shows efficient performance of the algorithm in terms of Probability of Detection PD, Probability of missed detection PM and the Probability of Error Pe in low and high noise variance environment.
{"title":"Multiscale Discrete Wavelet Transform based Efficient Energy Detection for Wideband Spectrum Sensing","authors":"Biji Rose, B. Arunadevi","doi":"10.1109/AISP53593.2022.9760625","DOIUrl":"https://doi.org/10.1109/AISP53593.2022.9760625","url":null,"abstract":"In a wireless radio environment of cognitive radio, spotting of vacant spectrum of Primary user demands more efficient technique. The edge detection of sub-bands of the received signal spectrum is one such efficient technique of spectrum sensing achieved by Discrete Wavelet Transform (DWT). In low noise variance, the DWT based technique has a better detection performance, but as noise variance increases, the performance degrades. In this paper, blind energy detection spectrum-sensing approach is proposed with Multiscale DWT. Here depending on the noise variance two modified forms of DWT are proposed. When noise variance is less DWT Modulus Maxima (DWTMM) and for high noise variance DWT Moving window ESPIT Method (DWTMEM). The simulation of the proposed algorithm, shows efficient performance of the algorithm in terms of Probability of Detection PD, Probability of missed detection PM and the Probability of Error Pe in low and high noise variance environment.","PeriodicalId":6793,"journal":{"name":"2022 2nd International Conference on Artificial Intelligence and Signal Processing (AISP)","volume":"12 1","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2022-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"87422602","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-02-12DOI: 10.1109/AISP53593.2022.9760651
Mrs.Latha S S, M. N, Mrs.Anusha Shettar
Document Verification using Blockchain Technology has a huge scope. With increasing documents generated every year, there is no systematic and simple way to verify the documents. This system could be used to the governments, organizations, employers and basically anybody who wants to verify that the given document is not forged. This could be used to verify all kinds of immutable records ranging from attendance records, birth certificates, graduation and academic credentials. The proposed system could be used by the government to construct a decentralized network to store and maintain record. This is also the best way to ensure that the documents exist in the state of their creation, that they are not tampered with by anyone. Motivated by this, we propose to develop a decentralized blockchain system using Ethereum that will serve as an application to authenticate the documents. An application will be installed to local systems in which the users will verify the documents. These local systems also known as “Nodes” or “Blocks”.Once the documents are added to blocks forming the blockchain, complex calculations are performed to find the unique hash for that particular document.This concept can be implemented through decentralized applications deployed on the blockchain. The blockchain that is intended to be used for the deployment process is the Ropsten Ethereum Network. Thus, the immutability of documents can be maintained, while providing a simple, yet secure way for authenticating/verifying documents.
{"title":"Block chain Based Framework for Document Verification","authors":"Mrs.Latha S S, M. N, Mrs.Anusha Shettar","doi":"10.1109/AISP53593.2022.9760651","DOIUrl":"https://doi.org/10.1109/AISP53593.2022.9760651","url":null,"abstract":"Document Verification using Blockchain Technology has a huge scope. With increasing documents generated every year, there is no systematic and simple way to verify the documents. This system could be used to the governments, organizations, employers and basically anybody who wants to verify that the given document is not forged. This could be used to verify all kinds of immutable records ranging from attendance records, birth certificates, graduation and academic credentials. The proposed system could be used by the government to construct a decentralized network to store and maintain record. This is also the best way to ensure that the documents exist in the state of their creation, that they are not tampered with by anyone. Motivated by this, we propose to develop a decentralized blockchain system using Ethereum that will serve as an application to authenticate the documents. An application will be installed to local systems in which the users will verify the documents. These local systems also known as “Nodes” or “Blocks”.Once the documents are added to blocks forming the blockchain, complex calculations are performed to find the unique hash for that particular document.This concept can be implemented through decentralized applications deployed on the blockchain. The blockchain that is intended to be used for the deployment process is the Ropsten Ethereum Network. Thus, the immutability of documents can be maintained, while providing a simple, yet secure way for authenticating/verifying documents.","PeriodicalId":6793,"journal":{"name":"2022 2nd International Conference on Artificial Intelligence and Signal Processing (AISP)","volume":"46 1","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2022-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83154646","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-02-12DOI: 10.1109/AISP53593.2022.9760599
Adal A. Alashban, Y. Alotaibi
In speech processing, identifying the speaker’s gender has been considered a topic of interest by many studies. Various approaches and methods have been proposed to detect the gender of a speaker with high accuracy. However, they are limited to isolated and specific languages. In this research, the speaker’s gender is classified from a mixed languages speech point of view, constituting six different languages using Bidirectional Long Short-Term Memory (BLSTM) network classifiers. Also, gender classification is performed using each specific language independently. The main aim of this approach is to tackle the effect of the language on speakers’ genders classification. Performance evaluation of the language effect on speaker gender classification is conducted on the open-source Mozilla datasets. We achieved an average gender classification accuracy of 90.42%, 97.42%, 82.44%, 98.39%, 100%, and 85.04% on Arabic, Chinese, English, French, Russian, and Spanish datasets, respectively. These results uncover some dependencies of speakers’ gender classification on the language.
在语音处理中，识别说话人的性别一直是许多研究感兴趣的话题。人们提出了各种方法和方法来检测说话人的性别，并且准确率很高。然而，它们仅限于孤立的和特定的语言。本研究从混合语言语音的角度出发，利用双向长短期记忆(Bidirectional Long - short - Memory, BLSTM)网络分类器对六种不同语言的说话人性别进行分类。此外，使用每种特定语言独立进行性别分类。这种方法的主要目的是解决语言对说话者性别分类的影响。在开源的Mozilla数据集上对语言对说话人性别分类的影响进行性能评估。在阿拉伯语、汉语、英语、法语、俄语和西班牙语数据集上，我们的平均性别分类准确率分别为90.42%、97.42%、82.44%、98.39%、100%和85.04%。这些结果揭示了说话者的性别分类与语言的一些依赖关系。
{"title":"Language Effect on Speaker Gender Classification Using Deep Learning","authors":"Adal A. Alashban, Y. Alotaibi","doi":"10.1109/AISP53593.2022.9760599","DOIUrl":"https://doi.org/10.1109/AISP53593.2022.9760599","url":null,"abstract":"In speech processing, identifying the speaker’s gender has been considered a topic of interest by many studies. Various approaches and methods have been proposed to detect the gender of a speaker with high accuracy. However, they are limited to isolated and specific languages. In this research, the speaker’s gender is classified from a mixed languages speech point of view, constituting six different languages using Bidirectional Long Short-Term Memory (BLSTM) network classifiers. Also, gender classification is performed using each specific language independently. The main aim of this approach is to tackle the effect of the language on speakers’ genders classification. Performance evaluation of the language effect on speaker gender classification is conducted on the open-source Mozilla datasets. We achieved an average gender classification accuracy of 90.42%, 97.42%, 82.44%, 98.39%, 100%, and 85.04% on Arabic, Chinese, English, French, Russian, and Spanish datasets, respectively. These results uncover some dependencies of speakers’ gender classification on the language.","PeriodicalId":6793,"journal":{"name":"2022 2nd International Conference on Artificial Intelligence and Signal Processing (AISP)","volume":"19 1","pages":"1-6"},"PeriodicalIF":0.0,"publicationDate":"2022-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"82568307","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-02-12DOI: 10.1109/AISP53593.2022.9760669
Chinnapapakkagari Sreenivasa Vikranth, B. Jagadeesh, Kanna Rakesh, Doriginti Mohammad, S. Krishna, Remya Ajai A S
In recent years, breast cancer has become one of the most prevalent kinds of cancer. Breast Ultrasound, Diagnostic Mammogram, Magnetic Resonance Imaging (MRI), and other imaging modalities are routinely used to diagnose breast cancer. Doctors make final judgments about treatments, drugs, and other matters based on biopsy results, which are regarded the standard diagnostic approach for cancer. However, this is a time-consuming process that also necessitates extensive pathologist training and expertise. Each pathology lab receives around 300-500 slides per day. This overburdens the pathologists and increases the misdiagnosis rate in the biopsy results. In order to provide timely error free results to the patients, the research community focuses more on the development of Computer Aided Diagnosis (CAD) System to assist pathologists to diagnose cancer. Recent developments in Deep Learning techniques made the CAD systems more effective in detecting breast cancer at an early stage with a great accuracy. In this paper, we present a CAD system that recognises histopathology images to diagnose breast cancer using a Convolutional Neural Network (CNN). DenseNet201, ResNet50 and MobileNetV2 are used in this work. These are trained and tested using the openly available BreakHis and BACH datasets. The datasets are subjected to binary and multi-class classifications. Accuracy, Precision, Recall, F1 Score, and AUC are all performance measures that are used to evaluate the model’s performance. For Binary classification, the model built using MobileNetV2 with Sigmoid as activation function displayed a higher accuracy of 97% - 98% and in the case of multi-class classification, again the model built using MobileNetV2 with Softmax as activation function displayed a higher accuracy of 91% - 92% for both Magnifican Independant (MI) and Magnification Dependant (MD) cases.
{"title":"Computer Assisted Diagnosis of Breast Cancer Using Histopathology Images and Convolutional Neural Networks","authors":"Chinnapapakkagari Sreenivasa Vikranth, B. Jagadeesh, Kanna Rakesh, Doriginti Mohammad, S. Krishna, Remya Ajai A S","doi":"10.1109/AISP53593.2022.9760669","DOIUrl":"https://doi.org/10.1109/AISP53593.2022.9760669","url":null,"abstract":"In recent years, breast cancer has become one of the most prevalent kinds of cancer. Breast Ultrasound, Diagnostic Mammogram, Magnetic Resonance Imaging (MRI), and other imaging modalities are routinely used to diagnose breast cancer. Doctors make final judgments about treatments, drugs, and other matters based on biopsy results, which are regarded the standard diagnostic approach for cancer. However, this is a time-consuming process that also necessitates extensive pathologist training and expertise. Each pathology lab receives around 300-500 slides per day. This overburdens the pathologists and increases the misdiagnosis rate in the biopsy results. In order to provide timely error free results to the patients, the research community focuses more on the development of Computer Aided Diagnosis (CAD) System to assist pathologists to diagnose cancer. Recent developments in Deep Learning techniques made the CAD systems more effective in detecting breast cancer at an early stage with a great accuracy. In this paper, we present a CAD system that recognises histopathology images to diagnose breast cancer using a Convolutional Neural Network (CNN). DenseNet201, ResNet50 and MobileNetV2 are used in this work. These are trained and tested using the openly available BreakHis and BACH datasets. The datasets are subjected to binary and multi-class classifications. Accuracy, Precision, Recall, F1 Score, and AUC are all performance measures that are used to evaluate the model’s performance. For Binary classification, the model built using MobileNetV2 with Sigmoid as activation function displayed a higher accuracy of 97% - 98% and in the case of multi-class classification, again the model built using MobileNetV2 with Softmax as activation function displayed a higher accuracy of 91% - 92% for both Magnifican Independant (MI) and Magnification Dependant (MD) cases.","PeriodicalId":6793,"journal":{"name":"2022 2nd International Conference on Artificial Intelligence and Signal Processing (AISP)","volume":"26 1","pages":"1-6"},"PeriodicalIF":0.0,"publicationDate":"2022-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"82631265","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-02-12DOI: 10.1109/AISP53593.2022.9760658
N. C. Dayananda Kumar, K. Suresh, R. Dinesh
Hand gesture recognition refers to identification of various hand postures which interprets the signs of non verbal communication. It finds various applications like Sign Language Recognition (SLR), Human Computer Interaction (HCI) for robotics control, 3D modeling etc., Efficiently recognizing the hand gestures in various complex background scenarios is still a challenging problem. This issue can be effectively addressed by using depth data as a additional cue along with RGB image. Depth refers to the distance between camera sensor and image scene, hence depth cues can be used in suppressing the complex backgrounds which are far away from the hand region. Depth can also be effectively used to handle the illumination issues. In this paper, we propose a two stage approach where first stage involves k-means algorithm based depth clustering and removal of the background region. In the later stage, the foreground filtered depth map is fused with RGB and the resultant RGB-D data is used for gesture recognition using Convolutional Neural Network (CNN) classification model. Experiments are conducted on OUHANDS datasets and the results are compared with well known existing methods. Experimental result shows that accuracy of 87.57 % can be achieved on OUHANDS test dataset using the proposed method.
{"title":"CNN based Static Hand Gesture Recognition using RGB-D Data","authors":"N. C. Dayananda Kumar, K. Suresh, R. Dinesh","doi":"10.1109/AISP53593.2022.9760658","DOIUrl":"https://doi.org/10.1109/AISP53593.2022.9760658","url":null,"abstract":"Hand gesture recognition refers to identification of various hand postures which interprets the signs of non verbal communication. It finds various applications like Sign Language Recognition (SLR), Human Computer Interaction (HCI) for robotics control, 3D modeling etc., Efficiently recognizing the hand gestures in various complex background scenarios is still a challenging problem. This issue can be effectively addressed by using depth data as a additional cue along with RGB image. Depth refers to the distance between camera sensor and image scene, hence depth cues can be used in suppressing the complex backgrounds which are far away from the hand region. Depth can also be effectively used to handle the illumination issues. In this paper, we propose a two stage approach where first stage involves k-means algorithm based depth clustering and removal of the background region. In the later stage, the foreground filtered depth map is fused with RGB and the resultant RGB-D data is used for gesture recognition using Convolutional Neural Network (CNN) classification model. Experiments are conducted on OUHANDS datasets and the results are compared with well known existing methods. Experimental result shows that accuracy of 87.57 % can be achieved on OUHANDS test dataset using the proposed method.","PeriodicalId":6793,"journal":{"name":"2022 2nd International Conference on Artificial Intelligence and Signal Processing (AISP)","volume":"5 1","pages":"1-6"},"PeriodicalIF":0.0,"publicationDate":"2022-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83480308","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-02-12DOI: 10.1109/AISP53593.2022.9760642
Ronali Madhusmita Sahoo, S. Padhy, Kumar Debasis
The task scheduling problem in a heterogeneous multiprocessor system is a challenging area of research. This article proposes a population-based metaheuristic algorithm called Modified Crow Search Optimization (MCSO) algorithm to solve the task scheduling problem. In this paper, the task scheduling problem is considered an optimization problem. The MCSO algorithm is used to find out the minimum makespan and the speedup of the task scheduling problem. The proposed algorithm is compared with some standard algorithms like Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Shuffled Frog Leaping Algorithm (SFLA), and Crow Search Optimization (CSO). Experimental results prove that the proposed algorithm outperforms all the above algorithms in minimizing the makespan.
{"title":"A New Dynamic Method of Multiprocessor Scheduling using Modified Crow Search Optimization","authors":"Ronali Madhusmita Sahoo, S. Padhy, Kumar Debasis","doi":"10.1109/AISP53593.2022.9760642","DOIUrl":"https://doi.org/10.1109/AISP53593.2022.9760642","url":null,"abstract":"The task scheduling problem in a heterogeneous multiprocessor system is a challenging area of research. This article proposes a population-based metaheuristic algorithm called Modified Crow Search Optimization (MCSO) algorithm to solve the task scheduling problem. In this paper, the task scheduling problem is considered an optimization problem. The MCSO algorithm is used to find out the minimum makespan and the speedup of the task scheduling problem. The proposed algorithm is compared with some standard algorithms like Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Shuffled Frog Leaping Algorithm (SFLA), and Crow Search Optimization (CSO). Experimental results prove that the proposed algorithm outperforms all the above algorithms in minimizing the makespan.","PeriodicalId":6793,"journal":{"name":"2022 2nd International Conference on Artificial Intelligence and Signal Processing (AISP)","volume":"16 1","pages":"1-6"},"PeriodicalIF":0.0,"publicationDate":"2022-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73194586","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：